Technology due diligence: a checklist for checking the sales deck against the code

Technology due diligence tests whether the software, security and team can carry the deal. A request checklist, and when to bring in an outside technical expert.

Instant Expert EditorialPublished Updated 6 min read

Technology due diligence checks whether a company's software, infrastructure, security and engineering team can support the plan behind an acquisition or investment. Put simply: does the product work the way the sales deck says, and what problems do you inherit once you own it? A request list and a few hours with the engineering team cover most of it. Bring in an outside technical expert when the technology is the main reason for the deal, when nobody on your side has built something similar, or when the findings will change the price.

Turn every seller claim into a question

Here is a made-up example. You are investing in a clinic scheduling software company with 12 engineers and a six-year-old codebase. The seller says the platform can handle five times today's customers without a rebuild, that security is "SOC 2 compliant" and that the team is stable. Each claim maps to a section of the checklist below. (The one in quotation marks deserves the closest reading.)

The technology due diligence checklist

Architecture and scale

  • A current system diagram: services, databases, third-party dependencies and where each runs
  • Monthly hosting costs for the last year, and what drives them
  • Load today versus the claimed headroom, and what broke at the last traffic spike
  • Single points of failure, and any core technology near end of life. U.S. banking regulators' vendor guidance lists "end of life issues with the software programming language, computer platform, or data storage technologies" as a resilience risk worth checking (Interagency Guidance on Third-Party Relationships)

Code and delivery

  • Read access to the main repositories, or a supervised walkthrough if the seller won't grant access before signing
  • How changes reach production: tests, reviews, deployment and rollback
  • Known technical debt, with the team's own estimate of the cost to fix it
  • Which parts of the code only one person understands

Security

  • Recent penetration test and vulnerability scan results, and what was fixed afterward
  • Multi-factor authentication, encryption and source code access management, all named as items to check in the same regulators' guidance
  • How the team finds and responds to vulnerabilities after release. NIST's Secure Software Development Framework groups secure development into four areas (prepare the organization, protect the software, produce well-secured software, respond to vulnerabilities). Use it for vocabulary; NIST says it is not meant to be a checklist to follow
  • Any SOC report, read in full. SOC 2 reports cover controls relevant to security, availability, processing integrity, confidentiality or privacy, and CPAs issue them (AICPA SOC suite). Check which systems and period it covers and what exceptions it lists. The AICPA's page currently carries a notice that it is looking into allegations about one SOC compliance vendor's practices, one more reason to read the report instead of trusting a badge

Open source and licenses

  • A software bill of materials (SBOM). CISA describes an SBOM as a nested inventory, "a list of ingredients that make up software components" (CISA SBOM)
  • The licenses of those components. The SPDX License List gives standard short identifiers for common open-source licenses, which makes an inventory easier to scan. Ask your attorney to review any license that places obligations on how you distribute the software

Data

  • What personal or sensitive data the product stores, where and for how long
  • Backups, and when a restore was last tested (until someone restores it, a backup is mostly a hope)

Team

  • The key engineers, their tenure and what happens if one leaves
  • The hiring the growth case assumes, and whether it is realistic

Bring in an outside expert when the technology is the thesis

Get outside help when any of these is true:

  • The technology is the reason for the deal. If you are paying for a proprietary system, its value depends on it being as good and as defensible as claimed.
  • Nobody on your side has built something similar. A deal team can read a diagram, but judging whether a six-year-old scheduling engine can scale five times over takes someone who has done it.
  • The domain is specialized. Payments, healthcare data, machine learning systems and embedded hardware each have failure modes a generalist may miss.
  • A finding would change the price. If a rebuild might be needed, get an estimate from someone independent of the seller.

That help usually comes in one of two forms:

  1. A full review. A specialist firm or senior engineer reviews the code and systems under a confidentiality agreement, with the seller's permission. It is the thorough option and the slowest.
  2. Short calls with practitioners. Before or alongside a full review, ask engineers and IT leaders who have run similar systems what usually goes wrong at this scale, how long the proposed migration tends to take, and what to ask the seller's team. Keep the seller's confidential materials out of these calls unless the seller has agreed.

How to choose the right expert covers matching someone's experience to your question. If you need someone to stay involved after the deal, consider a fractional CIO.

Write findings with a cost attached

For each area, list what you checked, what you found, what it would cost to fix, and what you could not check. "Two services run on a database version past vendor support; upgrade estimated at six to eight engineer-weeks by the team, not yet reviewed independently" is useful. "Some technical debt" is true of every company ever, so it tells the deal team nothing.

Technology findings feed into the wider due diligence checklist, and vendors the product depends on belong in vendor due diligence.

Your next step

Send the seller the request list above, trimmed to the claims your price depends on. Write down which of those claims you are not qualified to judge.

For short calls, describe the person, such as "engineering leads who have scaled a B2B scheduling product," and Instant Expert finds people who match and their work email. A 30- or 45-minute call is usually enough to test one claim, such as whether a migration the seller calls nearly done sounds nearly done to someone who has run one. You set what each person receives and our fee is added on top, so a $200 offer costs $250, charged only when they book. The directory pages for engineering professionals in enterprise SaaS and IT professionals in cybersecurity are a place to start.

Your note is what the engineer reads before deciding, so say who you are and what you won't ask: nothing confidential about the target or about their own current or former employer. They can book, decline, or reply with the price they want, and a counteroffer above your offer comes to you to accept or decline. We don't return the private email addresses we find, so the invitation is how you reach them.